arm64: smp: Support non-SGIs for IPIs

The arm64 arch has relied so far on GIC architectural software
generated interrupt (SGIs) to handle IPIs. Those are per-cpu
software generated interrupts.

arm64 architecture code that allocates the IPIs virtual IRQs and
IRQ descriptors was written accordingly.

On GICv5 systems, IPIs are implemented using LPIs that are not
per-cpu interrupts - they are just normal routable IRQs.

Add arch code to set-up IPIs on systems where they are handled
using normal routable IRQs.

For those systems, force the IRQ affinity (and make it immutable)
to the cpu a given IRQ was assigned to.

Signed-off-by: Timothy Hayes <[email protected]>
[lpieralisi: changed affinity set-up, log]
Signed-off-by: Lorenzo Pieralisi <[email protected]>
Cc: Will Deacon <[email protected]>
Cc: Catalin Marinas <[email protected]>
Acked-by: Catalin Marinas <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Marc Zyngier <[email protected]>

Author: Marc Zyngier

arch/arm64/include/asm/smp.h

@@ -53,7 +53,12 @@ extern void smp_init_cpus(void);
 /*
  * Register IPI interrupts with the arch SMP code
  */
-extern void set_smp_ipi_range(int ipi_base, int nr_ipi);
+extern void set_smp_ipi_range_percpu(int ipi_base, int nr_ipi, int ncpus);
+
+static inline void set_smp_ipi_range(int ipi_base, int n)
+{
+	set_smp_ipi_range_percpu(ipi_base, n, 0);
+}
 
 /*
  * Called from the secondary holding pen, this is the secondary CPU entry point.

arch/arm64/kernel/smp.c

@@ -83,7 +83,16 @@ enum ipi_msg_type {
 
 static int ipi_irq_base __ro_after_init;
 static int nr_ipi __ro_after_init = NR_IPI;
-static struct irq_desc *ipi_desc[MAX_IPI] __ro_after_init;
+
+struct ipi_descs {
+	struct irq_desc *descs[MAX_IPI];
+};
+
+static DEFINE_PER_CPU_READ_MOSTLY(struct ipi_descs, pcpu_ipi_desc);
+
+#define get_ipi_desc(__cpu, __ipi) (per_cpu_ptr(&pcpu_ipi_desc, __cpu)->descs[__ipi])
+
+static bool percpu_ipi_descs __ro_after_init;
 
 static bool crash_stop;
 
@@ -844,7 +853,7 @@ int arch_show_interrupts(struct seq_file *p, int prec)
 		seq_printf(p, "%*s%u:%s", prec - 1, "IPI", i,
 			   prec >= 4 ? " " : "");
 		for_each_online_cpu(cpu)
-			seq_printf(p, "%10u ", irq_desc_kstat_cpu(ipi_desc[i], cpu));
+			seq_printf(p, "%10u ", irq_desc_kstat_cpu(get_ipi_desc(cpu, i), cpu));
 		seq_printf(p, "      %s\n", ipi_types[i]);
 	}
 
@@ -917,9 +926,20 @@ static void __noreturn ipi_cpu_crash_stop(unsigned int cpu, struct pt_regs *regs
 #endif
 }
 
+static void arm64_send_ipi(const cpumask_t *mask, unsigned int nr)
+{
+	unsigned int cpu;
+
+	if (!percpu_ipi_descs)
+		__ipi_send_mask(get_ipi_desc(0, nr), mask);
+	else
+		for_each_cpu(cpu, mask)
+			__ipi_send_single(get_ipi_desc(cpu, nr), cpu);
+}
+
 static void arm64_backtrace_ipi(cpumask_t *mask)
 {
-	__ipi_send_mask(ipi_desc[IPI_CPU_BACKTRACE], mask);
+	arm64_send_ipi(mask, IPI_CPU_BACKTRACE);
 }
 
 void arch_trigger_cpumask_backtrace(const cpumask_t *mask, int exclude_cpu)
@@ -944,7 +964,7 @@ void kgdb_roundup_cpus(void)
 		if (cpu == this_cpu)
 			continue;
 
-		__ipi_send_single(ipi_desc[IPI_KGDB_ROUNDUP], cpu);
+		__ipi_send_single(get_ipi_desc(cpu, IPI_KGDB_ROUNDUP), cpu);
 	}
 }
 #endif
@@ -1013,14 +1033,16 @@ static void do_handle_IPI(int ipinr)
 
 static irqreturn_t ipi_handler(int irq, void *data)
 {
-	do_handle_IPI(irq - ipi_irq_base);
+	unsigned int ipi = (irq - ipi_irq_base) % nr_ipi;
+
+	do_handle_IPI(ipi);
 	return IRQ_HANDLED;
 }
 
 static void smp_cross_call(const struct cpumask *target, unsigned int ipinr)
 {
 	trace_ipi_raise(target, ipi_types[ipinr]);
-	__ipi_send_mask(ipi_desc[ipinr], target);
+	arm64_send_ipi(target, ipinr);
 }
 
 static bool ipi_should_be_nmi(enum ipi_msg_type ipi)
@@ -1046,11 +1068,15 @@ static void ipi_setup(int cpu)
 		return;
 
 	for (i = 0; i < nr_ipi; i++) {
-		if (ipi_should_be_nmi(i)) {
-			prepare_percpu_nmi(ipi_irq_base + i);
-			enable_percpu_nmi(ipi_irq_base + i, 0);
+		if (!percpu_ipi_descs) {
+			if (ipi_should_be_nmi(i)) {
+				prepare_percpu_nmi(ipi_irq_base + i);
+				enable_percpu_nmi(ipi_irq_base + i, 0);
+			} else {
+				enable_percpu_irq(ipi_irq_base + i, 0);
+			}
 		} else {
-			enable_percpu_irq(ipi_irq_base + i, 0);
+			enable_irq(irq_desc_get_irq(get_ipi_desc(cpu, i)));
 		}
 	}
 }
@@ -1064,44 +1090,77 @@ static void ipi_teardown(int cpu)
 		return;
 
 	for (i = 0; i < nr_ipi; i++) {
-		if (ipi_should_be_nmi(i)) {
-			disable_percpu_nmi(ipi_irq_base + i);
-			teardown_percpu_nmi(ipi_irq_base + i);
+		if (!percpu_ipi_descs) {
+			if (ipi_should_be_nmi(i)) {
+				disable_percpu_nmi(ipi_irq_base + i);
+				teardown_percpu_nmi(ipi_irq_base + i);
+			} else {
+				disable_percpu_irq(ipi_irq_base + i);
+			}
 		} else {
-			disable_percpu_irq(ipi_irq_base + i);
+			disable_irq(irq_desc_get_irq(get_ipi_desc(cpu, i)));
 		}
 	}
 }
 #endif
 
-void __init set_smp_ipi_range(int ipi_base, int n)
+static void ipi_setup_sgi(int ipi)
 {
-	int i;
+	int err, irq, cpu;
 
-	WARN_ON(n < MAX_IPI);
-	nr_ipi = min(n, MAX_IPI);
+	irq = ipi_irq_base + ipi;
 
-	for (i = 0; i < nr_ipi; i++) {
-		int err;
+	if (ipi_should_be_nmi(irq)) {
+		err = request_percpu_nmi(irq, ipi_handler, "IPI", &irq_stat);
+		WARN(err, "Could not request IRQ %d as NMI, err=%d\n", irq, err);
+	} else {
+		err = request_percpu_irq(irq, ipi_handler, "IPI", &irq_stat);
+		WARN(err, "Could not request IRQ %d as IRQ, err=%d\n", irq, err);
+	}
 
-		if (ipi_should_be_nmi(i)) {
-			err = request_percpu_nmi(ipi_base + i, ipi_handler,
-						 "IPI", &irq_stat);
-			WARN(err, "Could not request IPI %d as NMI, err=%d\n",
-			     i, err);
-		} else {
-			err = request_percpu_irq(ipi_base + i, ipi_handler,
-						 "IPI", &irq_stat);
-			WARN(err, "Could not request IPI %d as IRQ, err=%d\n",
-			     i, err);
-		}
+	for_each_possible_cpu(cpu)
+		get_ipi_desc(cpu, ipi) = irq_to_desc(irq);
+
+	irq_set_status_flags(irq, IRQ_HIDDEN);
+}
+
+static void ipi_setup_lpi(int ipi, int ncpus)
+{
+	for (int cpu = 0; cpu < ncpus; cpu++) {
+		int err, irq;
 
-		ipi_desc[i] = irq_to_desc(ipi_base + i);
-		irq_set_status_flags(ipi_base + i, IRQ_HIDDEN);
+		irq = ipi_irq_base + (cpu * nr_ipi) + ipi;
+
+		err = irq_force_affinity(irq, cpumask_of(cpu));
+		WARN(err, "Could not force affinity IRQ %d, err=%d\n", irq, err);
+
+		err = request_irq(irq, ipi_handler, IRQF_NO_AUTOEN, "IPI",
+				  NULL);
+		WARN(err, "Could not request IRQ %d, err=%d\n", irq, err);
+
+		irq_set_status_flags(irq, (IRQ_HIDDEN | IRQ_NO_BALANCING_MASK));
+
+		get_ipi_desc(cpu, ipi) = irq_to_desc(irq);
 	}
+}
 
+void __init set_smp_ipi_range_percpu(int ipi_base, int n, int ncpus)
+{
+	int i;
+
+	WARN_ON(n < MAX_IPI);
+	nr_ipi = min(n, MAX_IPI);
+
+	percpu_ipi_descs = !!ncpus;
 	ipi_irq_base = ipi_base;
 
+	for (i = 0; i < nr_ipi; i++) {
+		if (!percpu_ipi_descs)
+			ipi_setup_sgi(i);
+		else
+			ipi_setup_lpi(i, ncpus);
+	}
+
 	/* Setup the boot CPU immediately */
 	ipi_setup(smp_processor_id());
 }